Method of controlling ventilation and chilling systems to conserve energy in commercial buildings

Government funding

None FIELD OF USE

The invention is useful in energy management, and more particularly in the field of energy management in commercial buildings.

Background

Energy use analysis in commercial buildings has been performed for many years by a number of software simulation tools which seek to predict the comfort levels of buildings while estimating the energy use. The underlying principles of these tools concentrate on thermal properties of individual elements of the building itself, such as wall panels, windows, etc. The complexity and level of detail required to accurately simulate a commercial building often makes its' use prohibitive. The accuracy of such models has also been called into question in the research material. Following the construction and occupation of a new commercial building, the installed plant, such as boilers and air conditioning equipment, whose function is to provide suitable occupant comfort, is usually controlled by a building management system (BMS).

Through practical experience within the construction industry, it has become known that this plant is often over-sized and the use of the plant is often excessive. Common examples of this include plant operating for significantly longer than required including unoccupied weekends, heating and cooling simultaneously operating in the same areas due to construction or control strategy problems and issues with overheating and the use of cooling to compensate. Where the common problem of overheating occurs, the building envelope is quite efficient in dumping excess heat by radiation. In a similar manner, where buildings are over-cooled in summer, buildings are very effective in absorbing heat from the external environment to compensate. The utilization of this plant is not normally matched to the building envelope in which it operates and it is the intention to show how the method described in this document can help with this matching process.

Publication number 2013-0304269 A1 and publication number 2015-0198961 A1 teach a series of methods developed to provide a high-level view of thermal performance in a commercial building. This view is quick to implement and easily understood by facilities and maintenance staff. The methods facilitate a better understanding of the thermal performance of a building envelope, as constructed, and the interaction between this envelope and the building's heating and cooling plant, as installed. The thermal performance of the building envelope and how it interacts with the plant has been expressed as a series of time lags and profiles which are functions of external air temperature and solar activity. External temperature remains the most influential of the external weather parameters on energy usage. The lags and profiles have been developed to be derived from data which is readily available within modern conventional buildings.

Brief summary of the invention

Following teachings in publication number 2013-0304269 A1, where the derivation of a building's natural thermal lag was presented, and publication number 2015-0198961 A1 where a less data intensive method to calculate the natural thermal lag was derived, the following is an explanation of how selective ventilation of a building can be used to reduce energy consumption, particularly in spring, summer and autumn. Fresh air is a requirement in all commercial buildings, and to ensure fresh air reaches all areas, the air is usually forced by fan power through air-handling units. Since fresh air is required all-year-round, the air is tempered to ensure it is delivered at a temperature suitable for the occupants. This means the air is often heated in winter and cooled in summer. The thermal load to provide heated or cooled fresh air at a rate of perhaps 9 liters/sec per occupant can be very substantial. (see Chartered Institute of Building Service Engineers, CIBSE—Guide A Environmental Design 2015). The objective is to ensure that warm air is passing through the building in winter and cool air in summer. At certain times of year (spring and autumn), it is possible to use outside air in its passive or untempered state, since it is just at the correct temperature for internal use. It is also possible to use overnight air temperatures to pre-cool buildings in summer. Based on a weather forecast, this specification describes how newly derived thermal profiles facilitate the accurate prediction of when these useful energy saving opportunities can be applied.

In the inventive method taught herein, two important thermal profiles are introduced: the first which predicts the rate at which overnight forced ventilation (Overnight Forced Ventilation Rate or OFVR) cools open spaces in a commercial building as a function of external temperature, and the second which predicts when outside air may be suitable for direct daytime use in place of tempered (heated or cooled) fresh air (Daytime Forced Air Replacement or DFAR). The specification also makes use of two earlier defined thermal profiles: night-time natural cool-down profile slope NNCPS (defined in SHIEL006—US pub. no. US-2016-0195887-A1) and Solar Gain Rate SGR (defined in SHIEL008). OFVR is likely to be an option for building operations during spring, summer and autumn. DFAR is likely to be an option during spring and autumn.

Overnight Forced Ventilation Rate

The Overnight Forced Ventilation Rate (OFVR) represents, over the course of one night during suitable times of year, but probably late spring, summer or early autumn, a statistical relationship describing how a series of internal space temperatures vary with the real-time external air temperature on a 15-minute interval basis, when using untempered forced ventilation. The purpose of this statistical relationship or model is to facilitate the prediction of when the outside air temperature might be suitable for overnight cooling of spaces within the commercial building, particularly during warm summer months. Cooling is usually delivered to these spaces by airflow which has been lowered in temperature by passing the air over a series of chilled water filled coils. Since chilling is usually electrically powered and expensive to run, being able to predict when this cooling function can be delivered by outside untempered air can potentially deliver substantial energy reduction.

The observed fall of internal space temperature during overnight periods when the external air temperature is suitably low, happens at the same time as the night-time natural cool-down occurs. For this reason, the cooling effects of the NNCPS must be separately accounted for. This is done by rate vector separation using the well-known trapezoidal or triangular methods. The separation of these two cooling influences facilitates the further prediction of the amount of required external air (the flowrate or fan speed) required to cool down the internal space.

The time of year when this energy reduction opportunity is available will depend in major part on the geographical location of the building.

Day-Time Forced Air Replacement

The day-time forced air replacement (DFAR) is a statistical model derived from observed data which helps to predict when suitable external air (temperature and relative humidity) is available to be used in place of chilled or tempered air during the day and/or during occupied hours. In other words, chilling may not be required while these favorable external environmental conditions exist. DFAR has been found to depend on the real-time external air temperature and the rate at which the building responds to solar gain (expressed in the Solar Gain Rate or SGR in SHIEL008). In most commercial buildings, from practical experience, even when these external environmental conditions exist, a chiller load persists which results in wasteful energy consumption.

While the forecast of external temperatures is used in the DFAR model to determine the extent and when external air can be used to cool a building during occupied hours, the direct effects of solar gain must also be taken into account. For this reason, the method used to derive the Solar Gain Rate in SHIEL008 is used to isolate the effects of the solar gain on internal space temperatures and it is then possible to examine both suitable or sensible external untempered air cooling and solar gain as two conflicting effects on internal space temperature, the combined result being possible to predict.

This invention teaches a method to reduce the thermal energy used in a commercial building by use of thermal parameters which are derived from readily-available data both internal and external to the building. By deriving a statistical relationship for each of the OFVR and DFAR, based on the weather forecast, it is possible to determine if, when and for how long the mechanical cooling system can be turned off or disabled from supplying chilled water to the ventilation system, which in turn, supplies tempered fresh air to some of all of the building in question.

Brief description of drawings

The drawings listed are provided as an aid to understanding the invention

FIG. 1 Plot of test building B 1 natural thermal lag as a function of external temperature. External temperature is shown for reference

FIG. 2A Building space temperature profile prior to any efficiency interventions during the cooling season. Mechanical cooling enabled, occupancy and solar activity. A: Internal temperature rising appox 5 am; B: Ventilation system enabled 6.30 am; C: Occupancy and solar activity effects; D: Chiller pumps enabled 7.45 am

FIG. 2B Building space temperature profile post efficiency program showing the use of the ventilation system used to pre-cool the building. Based on the weather forecast of external air temperatures and derived OFVR models, A: Ventilation system enabled 4 am—variable frequency drives operate air fans between 80% and 50%; B: First occupancy 7 am; C: Suitable external air temperatures until 9 am (below 65 F); D: Chiller unit enabled as Tsp nears 72 F 10.15 am, chiller pumps remain disabled; E: Chiller pumps enabled 11 am

FIG. 3A Method 1 Inventive Process Steps 200 - 260

FIG. 3B Method 1 Inventive Process Steps 270 - 320

FIG. 3C Method 1 Inventive Process Steps 330 - 410

FIG. 3D Method 2 Inventive Process Steps 500 - 550

FIG. 3E Method 2 Inventive Process Steps 560 - 610

FIG. 3F Method 2 Inventive Process Steps 620 - 700

FIG. 4A Physical connections from building management system to plant and Modbus over IP

FIG. 4B —Inventive system connecting to the BMS Modbus over IP network

FIG. 5 B 1 agreed energy baseline data

FIG. 6 Building space temperature profile during early and late cooling season showing the use of external air to replace chilled water cooling while external air temperatures are below the required limit of 69 F. A: External air temperature falls below 69 F at 15:00, chiller pumps are disabled; B: Slight rise in external temperatures, chiller pumps are re-enabled at 16:30

FIG. 7 B 1 benchmark (BM) usage versus CIBSE usage ranges for heat and electricity

FIG. 8 B 1 thermal profile statistical models derived from on-site and observed data

FIG. 9 Total heat delivered to B 1 —over a four year period with the commencement of the energy efficiency program indicated by A.

FIG. 10 Total chilling delivered to B 1 —over a four year period

FIG. 11 Annual energy use outcomes for P 1 over the four year period

FIG. 12 Comparison of electricity and gas equivalent usage over calendar baseline year versus year 3 DETAILED DESCRIPTION OF A PREFERRED EMBODIMENT OF THE INVENTION

Introduction

The invention is a computer system capable of connecting directly to a commercial building management system. The purpose of the invention computer system is to provide improved control of plant operations to enable significant energy savings in commercial buildings while providing desirable occupant comfort levels.

This section describes the introduction of two new thermal profiles, the manner in which these profiles along with the natural thermal lag described in publication number 2013-0304269 A1 and publication number 2015-0198961 A1 can be applied to the control of plant in a particular building, and finally, the application of these concepts to an actual building and the energy reduction results. Two new building specific thermal profiles are introduced and are referred to as the Overnight Forced Ventilation Rate (OFVR) and Day-time Forced Air Replacement (DFAR).

Following publication number 2013-0304269 A1, where the derivation of a building's natural thermal lag was presented, and publication number 2015-0198961 A1 where a less data intensive method to calculate the natural thermal lag was presented, the following is an explanation of how the natural thermal lag, along with a number of important thermal profiles, can be combined to achieve automated optimization of energy usage in commercial buildings. The following sections recap on how the natural thermal lag is derived in publication number 2013-0304269 A1 and publication number 2015-0198961 A1, and also shows the derivations of the overnight forced ventilation rate and the daytime forced air replacement. Both of these thermal parameters have been shown to be closely correlated to the average daily external temperature. Two further thermal profiles: night-time natural cool-down profile slope (NNCPS) and the solar gain rate (SGR), introduced in SHIEL006 and SHIEL008, respectively, are also used in this specification. The time lag evident in the derivation of the NNCPS is closely correlated with the building's unique natural thermal lag.

Natural Thermal Lag

The derivation of the building-unique natural thermal lag can be summarized as follows (from publication number 2013-0304269 A1 and publication number 2015-0198961 A1):

The natural thermal lag (NTL) of a commercial building is a unique property which indicates how quickly the internal spaces of the building respond to changes in external temperature. The NTL can be derived as follows: a) using previously recorded data within said commercial building being 12 months of internal and external temperature data recorded at 15-minute intervals while the building was at rest, or in other words, the building was not in use, had no plant operating and experienced less than 1 hour of solar activity during the day in question (publication number 2013-0304269 A1). If internal temperature data is not available, the data used are energy consumption and external temperature data recorded at 15-minute intervals (publication number 2015-0198961 A1) b) deriving the natural thermal lag (NTL) of said commercial building by applying the sum of squares method (outlined in publication number 2013-0304269 A1) on the 12 months of internal and external temperature data only on days when the building was at rest, where each value of NTL is calculated according to:

LagIndex LW = .Math. i = 2 ⁢ p p ⁢ ⁢ ( T s i - T o i - LW ) 2 wherein LagIndex.sub.LW is a sum of squares particular to a range of external temperatures indicated by a value LW, p is a number of 15 minute observations examined, T.sub.S.sub. i is an internal space temperature at time period i, T.sub.o.sub. i-LW is an outside temperature at LW periods prior to time period i If internal temperature is not available, apply the building energy to external temperature data regression analysis method as follows: E .sub.i=β.sub.0 +B .sub.1( LT .sub.i).sub.k=0 . . . 8+ε.sub.i where E.sub.i represents average hourly energy usage for said building on day i, β.sub.0 represents a Y axis intercept of a linear relationship between energy and lagged temperature average, β.sub.1 represents a slope of a relationship between average hourly energy usage and a lagged temperature average (LT.sub.i).sub.k=0 . . . 8 for a day i and ranging over a period k from 0 to 8 hours prior to a building closing time, ε is estimated variation The particular index of lagged average external temperature during the winter yields the low point of NTL sinusoid, while the particular index of lagged average external temperature during the summer yields the high point of the NTL sinusoid. This yields an approximated NTL plot over the full year (publication number 2015-0198961 A1). c) Each NTL point (one for each day the building is at rest) can be plotted against the average external temperature recorded for that day. The relationship between the NTL and average daily external temperature can be established according to the regression equation: NTL .sub.i=β.sub.0−β.sub.1 T out.sub.i+ε.sub.i wherein NTL.sub.i is the natural thermal lag calculated on a particular day i β.sub.0 is the intercept of the linear relationship between NTL and the average daily external temperature Tout on the y-axis β.sub.1 is the slope of the linear relationship between NTL and the average daily external temperature Tout Tout.sub.i is the average daily external temperature calculated as the average of the 96 external temperature readings recorded during day i ε.sub.i is the variability in the linear relationship Once the particular relationship between NTL and daily average external temperature is established for said commercial building, the NTL can be estimated for any given average daily external temperature. Natural Thermal Lag Profile

Plotting the individual values of the natural thermal lag derived from data for each day the building is at-rest is indicated in FIG. 1 . From FIG. 1 , it is evident that the NTL is strongly related to the average daily external temperature. The strength of that relationship for this building can be examined by linear regression in which daily average outside temperature Tout.sub.i can be regressed against the observed NTL (based on results in publication number 2013-0304269 A1).

This relationship can be statistically modelled as a simple linear regression of: NTL .sub.i=β.sub.0−β.sub.1 T out.sub.i+ε.sub.i

The actual model derived for the test building B 1 is: NTL= 12.93−0.555 T out±1.9

The parametric statistics which define this relationship are shown as an extract from the Minitab statistical analysis package:

TABLE-US-00001 Regression Analysis: B1 NTL versus Average Tout The regression equation is NTL = 12.93 + 0.5546 Average Tout S = 0.851145 R-Sq = 91.7% R-Sq(adj) = 91.6% Analysis of Variance Source DF SS MS F P Regression 1 539.462 539.462 744.65 0.000 Error 67 48.538 0.724 Total 68 588.000

This particular NTL response curve in FIG. 1 is defined by the high and low points. The curve remains consistently sinusoidal in following the pattern of average external temperatures from year to year. Therefore, it follows that if the high and low points are known, the annual NTL response curve can be estimated.

In publication number 2015-0198961 A1, it has been shown how energy usage data of winter heating and summer cooling can be used to determine the optimum value of NTL for these seasons without any reference to internal temperature data.

In fact, these values of NTL for summer and winter represent the highest and lowest points of the sinusoid and therefore a method to determine the year-long NTL response for this building has been developed, based on energy usage and external temperature data alone.

This facilitates the simple estimation of the building's unique NTL to be used for energy efficiency purposes, in the event that rapid estimation is required or that a full year of internal space temperature data is unavailable.

The overnight forced ventilation rate and the day-time forced air replacement are now defined. They are useful in determining the best start times for overnight ventilation plant operation and day-time cooling replacement times based on the external temperature profile contained in a weather forecast. This section shows how these two thermal parameters can be applied to ventilation plant operation times and are therefore used to reduce energy consumption in commercial buildings.

Overnight Forced Ventilation Rate

The Overnight Forced Ventilation Rate (OFVR) represents, over the course of several days of measurements, during late spring, summer or early autumn, a statistical relationship describing how a series of internal space temperatures varies with the real-time external air temperature on a 15-minute interval basis, while the building is being supplied with untempered external air via the fresh air system. During this overnight period, with the fresh air system running, but no heating or chilling, at least two possible cooling influences are acting on the internal space temperatures:

the natural cool-down of the building's internal spaces due to a falling external temperature and subsequent heat loss through the building envelope, and

the cooling effect of the cooler external air being supplied to spaces within the building by the fresh air system. These two influences are of interest in this specification. The natural cool-down

is captured by the NNCPS while the cooling effect of cooler external air

includes both the effect of the NNCPS and the effect of the fresh air supply. To determine the effect of supplying cooler fresh air overnight alone, the effect of the NNCPS must be subtracted.

The purpose of this statistical relationship or model is to facilitate the prediction of when the outside air temperature and humidity might make it suitable for untempered air to be used to provide overnight cooling of spaces within the commercial building, particularly during warm summer months. Since cooling is usually electrically powered, being able to predict when this cooling can be delivered by outside untempered air can potentially deliver substantial energy reduction. The time of year when this energy reduction opportunity is available will depend in major part on the geographical location of the building.

During the cooling season, when overnight external temperature is generally below 65° F. and relative humidity is below 60%, the fresh air ventilation system is enabled at a chosen time, say 5 am. It is also on interest to determine how much heating occurs of the fresh air as it passes through ductwork in the spring and autumn. This heat gain can be expressed in time as the building's natural ventilation lag and is a function of both average hourly external air temperature and the volume of fresh air being forced through the building's ductwork. For most modern ventilation systems, the volume of air is controllable with variable frequency drives (VFD) fitted to the supply fans. If a VFD is not fitted, it is now low cost and small effort to fit such devices. The speed at which the fan needs to operate is related to the external air temperature and therefore the fan speed is included in the linear regression relationship. This allows the optimization of fan power usage as a function of external air temperature.

From the ventilation system start time, internal and external data are collected and a regression model is derived to show how the internal space temperature changes as a function of the difference between that space temperature and the realtime external temperature. The objective is to get the entire building to an internal space temperature of approximately 68° F. at the time of occupation. This will feel very comfortable at 8 am during the cooling season. This process is repeated for any overnight when suitable external environmental conditions persist.

The VFD speed starts at 100% at the ventilation system ON time and is gradually lowered each 30 minutes, perhaps by 10% reduction, depending on the lowering of internal space temperatures. On each successive night and over time, during the cooling season, a profile is created of the required average fan speed, given the prevailing external air temperature. With the recorded temperature data, a regression relationship is derived by using an equation: T .sub.SPi=β.sub.0−β.sub.1( T .sub.SPi −T out.sub.i)+ε.sub.i Eqn 1 wherein T.sub.SPi is the internal space temperature recorded at time period i β.sub.0 represents the intercept of the linear relationship between the internal space temperature and the difference between the internal space temperature and the external real-time temperature, on the y-axis β.sub.1 represents the slope of the relationship between the internal space temperature T.sub.SPi and the difference between that temperature and the real-time temperature Tout.sub.i at time period i Tout.sub.i is the value of real-time external temperature, observed for any given time period i ε represents the variability in the linear model

The slope of this linear relationship β.sub.1 is the OFVR for this particular overnight period. By deriving several values of OFVR, one for each day, and recording the average daily real-time external temperature during the same periods, a predictive relationship can be formed which indicates how the OFVR will vary as a function of daily average real-time external temperature and variable frequency drive speed. This yields a series of OFVR.sub.i=1 . . . N values for heating days 1 . . . N. This is shown in generalized form as follows: OFVR .sub.i=β.sub.0−β.sub.1 AT out.sub.i+β.sub.2 AVFD .sub.i+ε.sub.i Eqn 2 wherein OFVR.sub.i is the derived overnight forced ventilation rate on any given day i, on which the cooling system is operating β.sub.0 represents the intercept of the linear relationship between OFVR and daily average real-time external temperature on the y-axis and variable frequency drive speed on the z-axis β.sub.1 represents the slope in the Y-direction of the plane forming the multiple linear regression relationship between the two predictors ATout.sub.i and AVFD.sub.i and the response OFVR.sub.i β.sub.2 represents the slope in the Z-direction of the plane forming the multiple linear regression relationship between the two predictors ATout.sub.i and AVFD.sub.i and the response OFVR.sub.i ATout.sub.i represents the value of daily average real-time external temperature calculated for any given day i AVFD.sub.1 represents the value of daily average variable frequency drive fan speed calculated for any given day i ε represents the variability in the linear model

In SHIEL006, the night-time natural cool-down profile slope or NNCPS was described and applied to data collected in buildings during the heating season. The same general method can also be applied to data collected during the cooling season. Depending on geographical location, the cooling systems in buildings will probably run from mid-spring to mid-autumn. There may be a period during the height of summer when overnight natural cooling is not an option, given times when the overnight external temperature is higher than 65° F. This is particularly true in locations such as the Middle East and in certain southern Europe countries and US states. However, the principles described in this specification will certainly apply during spring and autumn, and these are sufficiently long periods to warrant their inclusion as part of an energy reduction program.

During periods when the external conditions are suitable, as described earlier, internal and external temperature data are recorded and the NNCPS algorithm is applied. This algorithm facilitates the prediction of how the overnight external temperature alone uniquely influences the internal space temperatures of any given building, while using the weather forecast of external temperatures. It is of particular interest to determine what the NNCPS model predicts will happen the internal space temperatures from the time the forced ventilation starts to the time of building occupation. The resulting thermal vector is simply subtracted from the equivalent OFVR thermal vector described above, to yield a resultant vector which is the effect of forced ventilation alone.

The two thermal vectors of NNCPS and OFVR must be derived from data recorded on two different nights. The separation of the vectors allows for the enhanced control of ventilation systems which have variable frequency drives installed. The cooler the external air, the slower the ventilation system can be run and it is possible to determine this slower fan speed in advance from the vectors which are both functions of external temperature. The objective is to get the building to the desired space temperature at the time of first occupation while having little or no chiller operation.

Invention Method 1

Method 1 steps are outlined in FIGS. 3A to 3C and are explained in the following section.

Method to determine suitable periods of overnight forced ventilation for space cooling prior to times of occupancy a) Determining [ 200 ] the building natural thermal lag by the means shown—these have shown in the preceding sections. Two methods exist and which one is used is determined by the data available. The methods to derive the natural thermal lag are more fully explained in U.S. Pat. No. 8,977,405 and in U.S. Pat. No. 9,317,026. b) Selecting [ 210 ] a suitable open plan area or space within a selected commercial building or a series of suitable open spaces in which to observe the space temperature(s); c) Determining [ 220 ] the internal building space set-point for the current cooling season. This is usually set at approximately 70-72° F. This is simply read off the building management system computer screen d) Determining [ 230 ] suitable (from weather forecast) overnight periods when the external temperature is generally below 65° F. with relative humidity of less than 60%. Enable the operation of the ventilation system at some agreed time, say 5 am e) Recording [ 240 ] the following data by observation during this ventilation system operating period in the said building: 1. space temperature(s) for the chosen open plan location(s) in 15 minute intervals from the time of ventilation system start-up until the space temperatures reach an average of 68° F. 2. simultaneous real-time external temperature in 15 minute intervals 3. relative humidity in 15 minute intervals to ensure no higher than 60% 4. lowering variable frequency drive speed every 30 minutes over the period from ventilation system start to time of occupation—this is averaged and recorded f) Deriving [ 250 ], using this recorded data when forced ventilation is enabled, a regression model to show how the internal space temperature changes as a function of the difference between that space temperature and the real-time external temperature for each cooling day using an equation: T .sub.SPi=β.sub.0−β.sub.1( T .sub.SPi −T out.sub.i)+ε.sub.i Eqn 1 wherein T.sub.SPi is the internal space temperature recorded at time period i β.sub.0 represents the intercept of the linear relationship between the internal space temperature and the difference between the internal space temperature and the external real-time temperature, on the y-axis β.sub.1 represents the slope of the relationship between the internal space temperature T.sub.SPi and the difference between that temperature and the real-time temperature Tout.sub.i at time period i Tout.sub.i is the value of real-time external temperature, observed for any given time period i ε represents the variability in the linear model g) Recording [ 260 ] the slope of Eqn 1 β.sub.1 is the OFVR for this particular overnight period. For each overnight observed, a predictive relationship can be formed which indicates how the OFVR will vary as a function of daily average real-time external temperature and average variable frequency drive speed. This yields a series of OFVR.sub.i=1 . . . N values for heating days 1 . . . N. This is shown in generalized form as follows: OFVR .sub.i=β.sub.0−β.sub.1 AT out.sub.i+β.sub.2 AVFD .sub.i+ε.sub.i Eqn 2 wherein OFVR.sub.i is the derived overnight forced ventilation rate on any given day i, on which the cooling system is operating β.sub.0 represents the intercept of the linear relationship between OFVR and daily average real-time external temperature on the y-axis and variable frequency drive speed on the z-axis β.sub.1 represents the slope in the Y-direction of the plane forming the multiple linear regression relationship between the two predictors ATout.sub.i and AVFD.sub.i and the response OFVR.sub.i β.sub.2 represents the slope in the Z-direction of the plane forming the multiple linear regression relationship between the two predictors ATout.sub.i and AVFD.sub.i and the response OFVR.sub.i ATout.sub.i represents the value of daily average real-time external temperature calculated for any given day i AVFD.sub.i represents the value of daily average variable frequency drive fan speed calculated for any given day i ε represents the variability in the linear model h) Recording [ 270 ] the following data from the building management system computer screens and physically verified during the night-time natural cool-down phase in the evening for said building by recording: 1. cooling plant shut-down time 2. space temperature(s) for the chosen open plan location(s) at this shut-down time (usually 70-72° F.) 3. Space temperature(s) for the chosen open plan location(s) at the time when cooling usually starts the following morning 4. external temperature data in 15 minute intervals i) Deriving [ 280 ], using this recorded data, a regression model to show how the internal space temperature changes as a function of the difference between that space temperature and the lagged external temperature for each cooling day using an equation: T .sub.SPi=β.sub.0−β.sub.1( T .sub.SPi−Lagged T out.sub.i)+ε.sub.i Eqn 3 wherein T.sub.SPi is the internal space temperature recorded at time period i β.sub.0 represents the intercept of the linear relationship between the internal space temperature and the difference between the internal space temperature and the external lagged temperature, as guided by the NTL for this time of year, on the y-axis represents the slope of the relationship between the internal space temperature T.sub.SPi and the difference between that temperature and the external lagged temperature LaggedTout.sub.i at time period i LaggedTout.sub.i is the value of lagged external temperature, as guided by the NTL for this time of year, observed for any given time period i ε represents the variability in the linear model j) Determining [ 290 ] the night natural cool-down profile slope (NNCPS) on days the cooling system is operating, to help estimate the starting point for the internal space temperature at cooling start time for each day on which the cooling is operating, repeat the process outlined in g), recording each average daily lagged external temperature and the slope of the regression relationship pertaining to that particular day, β.sub.1 or NNCPS. In this regression model (Eqn 3), the slope β.sub.1 will be referred to as the NNCPS.

This yields a series of NNCPS.sub.i=1 . . . N values for cooling days 1 . . . N. A relationship can be established which links the NNCPS to the average daily average lagged external temperature and this is shown in generalized form in Eqn 4: NNCPS .sub.i=β.sub.0−β.sub.1 A Lagged T out.sub.i+ε.sub.i Eqn 4 wherein NNCPS.sub.i is the derived night-time natural cool-down profile slope on any given day i, on which the cooling system is operating β.sub.0 represents the intercept of the linear relationship between NNCPS and daily average lagged external temperature as guided by the natural thermal lag on the y-axis β.sub.1 represents the slope of the relationship between NNCPS.sub.i and daily lagged average external temperature ALaggedTout.sub.i ALaggedTout.sub.i represents the value of daily average lagged external temperature guided by the natural thermal lag calculated for any given day i ε represents the variability in the linear model k) Using [ 300 ] the relationships formed in Eqn 2 and Eqn 4, take the predicted rise and/or fall of internal space temperature(s) during the period when the ventilation system is operational to form two resulting thermal vectors. Apply either the trapezoidal or triangular method of vector subtraction to yield a prediction of how the building space temperatures will respond due to overnight forced ventilation alone l) Gathering [ 310 ] the hourly weather forecast to include 15 minute predictions of external temperature for the following 8-12 hours, ensuring the forecast extends beyond the estimated natural thermal lag of the commercial building in question during the cooling season. Using this forecast in conjunction with Eqns 1-4 to predict a suitable start-up time and fan speed for the ventilation system to ensure correct space temperatures in the building at the time of first occupancy m) Commencing [ 320 ] ventilation system start-up at the predicted time n) Performing [ 330 ] a communication between the invention computer and the BMS using a protocol such as Modbus over IP to enable the ventilation system. For example, if the hex value of 0x1010 represents ‘Ventilation system ENABLE’ if placed in Modbus register 8056 , as agreed with the BMS programmer o) Writing [ 340 ] an agreed test count value into an agreed register to ensure the BMS knows the invention computer is present and functional p) Awaiting [ 350 ] the response from the BMS, to indicate to the invention computer that the BMS is responsive q) Placing [ 360 ] the 0x1010 data value into the agreed Modbus over IP protocol register at the calculated ventilation system on time r) Reading [ 370 ] the confirmation response from the BMS in another register to confirm to the invention computer that the instruction to enable the ventilation system has been received s) Responding [ 380 ] to this writing of digital data (0x1010) into this register ( 8056 ), the BMS will turn the ventilation system on t) Depending [ 390 ] on the results of the vector combination explained in step 300 , it has been determined that the ventilation system may be enabled for some period to avoid using the chiller. Once the building reaches occupation time, when solar activity, lighting and occupant effects may cause a rise in space temperature, the operation of the building's plant will return to normal set-points and schedules u) Recording [ 400 ] permanently, the observed 15-minute interval data for weather forecast, internal space temperatures and all other relevant data used in the above equations to facilitate more accuracy in the data regression models, to effectively allow for machine learning over time v) Repeating [ 410 ] steps d) 230 to u) 400 at an appropriate time, as calculated, to determine an optimum ventilation system early morning enable time during the cooling season. Day-Time Forced Air Replacement

During the early and late parts of the cooling season, it is generally accepted that tempered fresh air should be supplied just below the desired internal space temperature set-point. Anything cooler might be noticeably cold. This supply temperature can be lowered at the height of summer, given occupants may wish for a cooler temperature. If external air is forced through ductwork in any part of the cooling season, it will likely heat up. The extent to which this heat-up occurs, is very dependent on external air temperature and solar activity and how these affect the building's façade and internal surfaces. For the purpose of this specification, it is assumed that in early and late parts of the annual cooling season, daytime external air temperatures are often below 66° F. and this is certainly true in parts of the US and Europe. From practical experience, it is also assumed that during these periods, the ductwork heat-up of forced air will add between 3° F. and 5° F. to the air volume's temperature. This would imply that external untempered air with a temperature of less than or about 66° F. with relative humidity of less that 60% would be suitable to use directly in buildings where the desired internal space temperature is about 71-72° F.

The description continues in the full USPTO document.